1

Do incentive payments encourage innovation? A meta-analysis study

Presented by:Zahra Lotfi

Friedrich Schiller University

MAER-NET 2015

2

Outline

• Introduction

• Literature

• Method

• Results

• Conclusion

3

Introduction• Human capital has been considered as a key resource

in innovative organizations. Successful innovative organizations should know how to reward people (Gupta and Singhal, 1993).

• Reward systems are effective in fostering innovation: long-term perspective, autonomy and motivation to take risks.

• Which type of reward system? Variable payments?

• In variable payment, the payment of individuals are linked to firm performance, like bonus and stock option.

4

• Previous studies about the effect of variable pay on innovation:variable payments are proper stimulant for innovation activity create long-term commitment

variable payment can not encourage innovation focus on short term, reduce intrinsic motivation, autonomy and freedom.

• A systematic overview of studies that examine the relation between variable payments and innovation is needed

• Meta-analysis: summerizes and integrates the results of emprical studies

Introduction

5

Positive impact of variable compensation on innovation:

• Variable monetary reward considered as an effective motivational tool for improving innovation.

• Apply principal-agent theory: To align the interests of agents and principals, the agents‘ payment linked to firm performance.

• Regarding innovation activities, the payments of agents should be linked to innovation activity of firms rather than firm‘s financial performance (Balkin et al., 2000)

• Long-term payment such as stock option has been found to be effective in promoting innovation (Francis et al., 2010; Lerner and Wulf, 2007; Yanadori and Marler, 2006)

Literature

6

Positive impact of variable compensation on innovation:

• Long-term pay can encourage employees to focus on the firm‘s long-term success (Chang et al. 2015).

• Long-term pay decrease the fear of failure in executives (Francis et al., 2010).

• Stock option can prevent myopic decision of CEOs (Sanders and Hambrick 2007).

• Stock option can encourage CEOs to take risky decisions (Sanders and Hambrick 2007).

Literature

7

Negative impact of variable compensation on innovation:

• Variable compensation negatively affect innovation.

• Monetary compensation can undermine intrinsic motivation of people in interesting tasks (e.g. Amabile, 1998; Fehr and Gächter, 2001; Frey and Oberholzer-Gee, 1997; Lepper et al., 1973).

• When intrinsically motivated people are paid stock options or bonus for doing interesting tasks like innovation, they might lose their interests in what they are doing and only focus on the reward.

• When monetary rewards are perceived as controlling, people are under pressure to achieve specific goals so their intrinsic motivation get reduced in interesting tasks.

Literature

8

Negative impact of variable compensation on innovation:

• Multitasking problem Individuals focus on the tasks that increase firm’s share price and value in short run in order to increase their payment, so it’s unlikely that they invest in innovation activity.

• Top managers focus on the tasks that enhance the firm’s profit in the shortest way they can impress boards and increase their payment

• R&D employees increase only the numbers of patents to ehance their payment without considering the quality of innovation.

Literature

9

• Meta-regression analysis can review all existing empirical studies comprehensively and by aggregating the results of various studies it can provide more authentic estimates than the individual study.

• We considered previous empirical studies that examine the relation between variable pay and innovation.

• We searched Google scholar, Elsevier, Business Source Premier and Jstor databases for the combination of the specific keywords.

• Variable payment of managers, employees and R&D heads are included.

• Compensation: long-term and short-term compensation; Innovation: patent, R&D intensity and innovation performance

• The final sample consists of 43 studies that report 301 estimation of compensation-innovation relationship.

Method

10

• Meta-regression model:

Dependent variable: Effect size measures the strength and direction of compensation-innovation link; Partial correlation

Independent variable: effect size standard error

Method

dft

tr

2

1

)1( 22

df

rVr

𝑆𝐸 𝑖=√𝑉 𝑟

11

Reporting Bias analysis:• Reporting bias: the propensity in reporting statistically significant

results.

• Basically authors and journals prefer to publish significant results and results which are more in accord with theories (Card and Krueger, 1995)

• Reporting bias can be considered as a threat to an empirical inference and validity of policy implication that are drawn from empirical results.

• Meta-regression analysis (MRA) helps us to detect the publication bias and identify the precision effect regardless of bias (Stanley, 2008).

• Reporting bias can be distinguished by two tests: Funnel plot test and FAT test (Egger et al., 1997; Stanley and Doucouliagos, 2010)

Method

12

• Funnel Plot Test:

010

2030

4050

1/st

d_r

-.5 0 .5 1r

Method

13

FAT test (funnel asymmetry testing)

H0: reporting bias

H0: genuine effect between compensation and innovation that adjusted for reporting bias• Weighted Least Square (WLS) has been applied• Fixed effect and Random effect models• Fixed effect model: the weight assigned to each study is inverse of

effect size variance (within-studies variance)• Random effect model: weigh each study by inverse of effect size

variance (within-studies variance and between-studies variance)

Method

14

FAT test:

𝑡𝑖=𝛽1+𝛽0(1 /𝑆𝐸𝑖)+𝑒𝑖

FAT Test

Dependent variable: t-statistics (1) (2)

Sub-effect, fixed effect

Sub-effect, random-effect

Precision 0.072***(3.41)

0.07***(4.97)

Constant (Reporting bias)-0.46(-0.64)

-0.410(-1.14)

Number of observation 301 301

R-squared 0.037 0.073

Results

15

FAT test:

Table . FAT-PET-MAR Test

Dependent variable: t-statistics (1) (2)

US studies, fixed effect

Non US studies, fixed effect

Percision 0.08** 0.054*

(3.01) (2.02)

Constant (Reporting bias) -0.7 0. 41

(-0.80) (0.48)

Number of observation 233 68

R-squared 0.07 0.12

Results

16

Moderator Variables:– Theoretical aspect– Publication outlet : Journal, type of journal, impact factor of

Journal, publication year– Sample characteristics: US firms or non US firms, type of

industry, compensation of different groups – Methodological aspect: Methods of analysis, industry fixed

effect, time period fixed effect and time-lag effects– Data: Database, Types of data– Control variables: innovation specifications, compensation

specifications, time period of sample, Firm characteristics, CEO characteristics

Method

17

Dependent Variable: t-statistics

Specific Model Input-based measure of innovation

Output-based measureof innovation

precision 0.122***(5.14) 0.391***(4.25) -----

Crowding-out Theory 2.64*** (4.49) 2.612(0.57) 4.62 (1.60)

Finance -2.6*** (-3.51) -4.505 (-1.72) 1.66 (0.99)

Working paper -2.46*** (-3.56) -2.3 (-0.54) -----

Employees -2.24*(-2.48) 6.151 (0.45) 1.075 (0.27)

Cross section 1.639** (2.69) ----- -----

Innovation performance -3.246* (-3.14) ----- -----

Stock option ----- -2.683 (-1.35) -2.844***(-5.13)

Stock 1.945** (2.64) -0.87 (-0.58) 4.525*** (8.12)

Bonus ----- -3.338* (-2.15) -----

Constant -0.233 (-0.33) 246.3 (0.39) -509.8 (-1.34)

N.observation 273 113 128

18

• No evidence of publication bias can be found among the selected studies.

• There is positive relationship between incentive pay and innovation; However this association is weak.

• The variation in estimated association between compensation and innovation across studies is due to differences in some study characteristics.

• The variable compensation are more effective in managers rather than employees.

• The way innovation and compensation are measured affect the compensation-innovation link.

• There are differences in the effect of short and long-term compensation on input and output-based measure of innovation.

Conclusions

19

Thanks for your attention!

20

21